pygwtf.utils#

Functions#

complex_inner_product(h1, h2, inv_psd)

Module Contents#

pygwtf.utils.complex_inner_product(h1, h2, inv_psd)#

‘ Usual noise-weighted inner product between two time-series in the frequency domain. Base of every statistic.

NOTE: Assumes inv_psd is 1/psd* 4*df.

Parameters#

h1array_like or float

First time-series in the frequency domain.

h2array_like or float

Second time-series in the frequency domain.

inv_psdarray_like or float

Reciprocal (1/psd) of the power spectral density of the noise. Prefolded outside the kernel so the inner loop multiplies instead of dividing (fp64 division is far slower than multiply on GPU).

Returns#

array_like or float

Noise-weighted inner product of the two time-series.